Spokenly converts speech to punctuated text system-wide across macOS and iOS using local AI models for privacy-focused dictation. The app runs OpenAI Whisper and NVIDIA Parakeet models entirely on Apple Silicon with offline functionality that blocks all network requests. (Free with local models; $9.99/month for Pro)
The voice recognition engine supports over 100 languages with automatic language detection and smart prompts for grammar correction compatible with GPT-4, Claude, and Gemini. Agent Mode enables voice command control for macOS automation, app launching, and system shortcuts without typing.
All processing occurs locally on-device—voice data never leaves the Mac when using included Whisper and Parakeet models. Cloud processing options include immediate audio deletion after transcription. The app maintains full transcription history with audio playback and export capabilities for reference and archival purposes.
Subscription tiers include free unlimited usage with local models, free bring-your-own-key integration with OpenAI, Deepgram, Groq, Anthropic, and Google APIs, or $9.99/month Pro access for managed cloud models and priority support. A single subscription covers both macOS and iPhone devices.
Resource usage runs natively on M1, M2, M3, and M4 Apple Silicon processors. The app has accumulated 100,000+ active users with a 4.9/5 App Store rating across 300+ reviews.
System requirements: Apple Silicon Macs (M1 or newer) for optimal local model performance. Available for macOS and iOS.
Limitations: Requires Apple Silicon for local model functionality—Intel Macs may need cloud-based processing. Agent Mode automation features limited to macOS. Accuracy varies by accent, background noise, and language selection.
Alternatives: macOS built-in dictation, Whisper (command-line), Otter.ai, Dragon NaturallySpeaking.
Suitable for users who need extensive dictation across applications with strong privacy requirements. Best for multilingual users, writers, accessibility needs, remote workers documenting meetings, or anyone requiring offline voice-to-text without cloud dependencies.
